Agents and Artificial Life
Agents and Artificial Life
Agents and artificial life are, to me, one of the most interesting and exciting fields in computer science. Currently, my primary work is to create, and facilitate the creation of, agent-based models to help people understand the dynamics that arise in real-world systems like ecologies and societies.
My Projects in Agents and Artificial Life
My History with Agents and Artificial Life
My interest in artificial life started sometime around 1992, when I read Stephen Levy's book, "Artificial Life: Quest for a New Creation". My excitement for this field hasn't dissipated since then.
Additional books that influenced me are "Out of Control", by Kevin Kelly and, lately, "Creation" my Steve Grand. I maintain a Listmania! list of these types of books on Amazon.com.
My own explorations into artificial life started also around 1992, when I started to program genetic evolution simulators, Lindenmayer System programs that created graphical plant forms and musical patterns, and assorted other applications.
One of my most significant creations from the time was SITISEP, Structued If-Then Isomorphic Sensor-Effector Processor, which had a notion of artificial creatures that could understand their environment through pluggable sensors and effectors. For example, you could create a creature with a light sensor, and a rule that said "Move towards the light"; you could then swap the light sensor with a smell sensor, and the creature would "Move towards the smell". SITISEP evolved into SEPIA (Sensor-Effector Processor for Intelligent Automata). Some of the core parts of SEPIA evolved into BaseAgent. Yes, I conceived parts of this brand-new thing called BaseAgent eleven years ago, but today, BaseAgent's full potential can be realized - and, I've had a lot of new thoughts since then! BaseAgent, and the applications I develop with BaseAgent, will exhibit all of the promise that one would expect from eleven years of deep and active thought.
My motivations for creating BaseAgent are threefold:
- To create an easy-to-use programming API that others can use to create their own experiments in simulation of agent-based models and artificial life
- To facilitate the creation of an end-user application for exploring systems with interacting agents, such as the interactions in an ecology, or robots exploring other planets
- To help realize my own advances in artificial life, such as the emergence of group dynamics with an artificial society
As an undergraduate at WPI, my interests expanded into the field of agents. Whereas artificial life tries to model biological processes, or to explore life-as-it-can-be, agent systems deal primarily with solving problems through collaborating entities that have specialized knowledge and abilities -- but this is a very simplified definition of what agents do.
My senior thesis at WPI involved exploring agent-based system solving configuration problems, and I developed an application that lays out newspaper pages. The program created realistic and convincing layouts, and I even used the program to layout pages of the school newspaper.
I also did an indepdent study on "Models of Decentralized Organization", a very interdisciplinary study that examined systems without centralized leaders. Topics included flocking, memory, and collective robotics. One of the deliverables for this study was a series of web pages explaining decentralized systems; those pages are currently being reconstructed and will be enhanced with BaseAgent simulations, so be sure to check back for that sometime in Fall 2003!
I currently have a patent pending for an idea based on artificial life: Patent Application #20030005358, A decentralized, self-regulating system for automatically discovering optimal configurations in a failure-rich environment (one of two patents I currently have pending in the U.S. Patent Office - here's the other patent).
My current interests in artificial life are centered around cooperative intelligent behavior, sociobiology, and evolutionary ecology. This is a combination of cognitive science and artificial life, and includes topics like nest building, grouping, and predator/prey strategies.